MicrobialEcology

Comparison of Bacterial Communities in New EnglandSphagnum Bogs Using Terminal Restriction FragmentLength Polymorphism (T-RFLP)

Sergio E. Morales1, Paula J. Mouser2, Naomi Ward3,4, Stephen P. Hudman5, Nicholas J. Gotelli5,Donald S. Ross6 and Thomas A. Lewis1

(1) Department of Microbiology and Molecular Genetics, University of Vermont, 95 Carrigan Drive, Burlington, VT 05405, USA(2) Department of Civil and Environmental Engineering, University of Vermont, Burlington, VT 05405, USA(3) The Institute for Genomic Research, 9712 Medical Center Drive, Rockville, MD 20850, USA(4) Center of Marine Biotechnology, 701 E. Pratt St, Baltimore, MD 21202, USA(5) Department of Biology, University of Vermont, Burlington, VT 05405, USA(6) Department of Plant and Soil Science, University of Vermont, Burlington, VT 05405, USA

Received: 22 December 2004 / Accepted: 10 May 2005 / Online publication: 31 May 2006

Abstract

Wetlands are major sources of carbon dioxide, methane,and other greenhouse gases released during microbialdegradation. Despite the fact that decomposition ismainly driven by bacteria and fungi, little is knownabout the taxonomic diversity of bacterial communitiesin wetlands, particularly Sphagnum bogs. To explorebacterial community composition, 24 bogs in Vermontand Massachusetts were censused for bacterial diversityat the surface (oxic) and 1 m (anoxic) regions. Bacterialdiversity was characterized by a terminal restrictionfragment length (T-RFLP) fingerprinting technique anda cloning strategy that targeted the 16S rRNA gene.T-RFLP analysis revealed a high level of diversity, and acanonical correspondence analysis demonstrated markedsimilarity among bogs, but consistent differences betweensurface and subsurface assemblages. 16S rDNA sequencesderived from one of the sites showed high numbers ofclones belonging to the Deltaproteobacteria group.Several other phyla were represented, as well as twoCandidate Division-level taxonomic groups. These datasuggest that bog microbial communities are complex,possibly stratified, and similar among multiple sites.

Introduction

Sphagnum bogs store one-third of the earth’s carbon, andin New England they are responsible for 36% of methaneemissions [10]. In spite of the importance of micro-

organisms to the decomposition of peatland vegetation[13, 21, 39, 40] and thus their effect on nutrient release,there has been little study of the taxonomic diversity ofpeatland microbial communities.

Bacterial numbers are relatively low in peat com-pared to aerated soils [30]. Bacterial density in high-latitude peatlands, including fens and bogs, peaks inspring and fall coinciding with seasonal nutrient pulses[10]. Total cell counts in bogs are rather constant (108–109 mLj1), although metabolic activity can fluctuate[19, 41]. To date, bacterial diversity has been describedfrom one Sphagnum bog, using only cultivation-basedapproaches [41]. Peatlands are known to have a widevariety of microbes capable of growing and metabolizingover a wide range of pH and growth conditions.Williams and Crawford [41] noted that bacterial generain peatlands include Bacillus, Pseudomonas, Achromo-bacter, Cytophaga, Micrococcus, Chromobacterium, Clos-tridium, Streptomyces, Actinomyces, Mycobacterium,Micromonospora, and Nocardia. Such a diverse group ofbacteria suggests a complex ecological web, with mi-crobes specialized for many niches. Metabolic relation-ships, such as the one serving to maintain a controllingbalance over the release of methane (methanotrophs andmethanogens), have been indicated by targeted surveys[7, 10, 17]. Nevertheless, our knowledge of bacterialcommunities in bogs is limited to specific groups ofbacteria from studies mostly targeting individual sites. Toassess the ability of bogs to alter levels of greenhousegases and to understand the microbial processes involvedin them, we first need to address the lack of informationon bacterial composition in this type of wetland.Correspondence to: Thomas A. Lewis; E-mail: talewis@zoo.uvm.edu

DOI: 10.1007/s00248-005-0264-2 & Volume 52, 34–44 (2006) & * Springer Science+Business Media, Inc. 200634

In this study, we characterized bacterial diversitywithin and among ombrotrophic Sphagnum bogs by sam-pling 24 New England bogs using rDNA-based techni-ques [terminal restriction fragment length (T-RFLP)[1, 2, 25, 26, 27] and sequencing]. Within each bog, wesampled both surface and anoxic subsurface microbialcommunities. Based solely on their shared oligotrophicconditions and acidic nature, we hypothesized that bac-terial diversity might be relatively similar among bogs,but that surface and subsurface assemblages would betaxonomically distinct, reflecting the different abioticconditions in these two microhabitats. Although water-shed and land cover are additional variables potentiallyinfluencing community composition, the data supportedthe view that the studied sites are hydrologically isolatedunits (i.e., ombrotrophic) and that the potential variablesof watershed or land cover had little or no effect on thebacterial composition.

Methods

Sample Collection. Liquid samples, with varyingamounts of suspended particulate matter and thereforecontaining attached and unattached bacteria, werecollected for bacterial enumeration and T-RFLP analysisfrom 24 different bogs throughout New England (Table 1)

during a 2-and-1/2-week period in June of 2002. We alsocollected a single surface sample from Molly Bog, VT, inJune of 2001 to create a clone library. These bogs havebeen classified elsewhere [11, 18] as acidic peatlands(mean pH = 4.97, range 4.03–6.37) with an activeSphagnum layer extending to a depth of 40 cm, withtwo of them also described as having anoxic waterloggedconditions beneath that depth [10]. Sampling locationswithin bogs were standardized to include areas with thehighest density of Northern Pitcher Plants (Sarraceneapurpurea) at the mat surface [18]. The presence of thosecarnivorous plants serve as an indicator for levels of Fe3+,Na+, Clj, SO4

j [31], and general ombrotrophy [11].Two samples were taken from each site, one from the top15.24 cm of the peat mat and another from 1 m deep(representing oxic and anoxic zones, respectively).Analyses to establish the depth at which anoxic con-ditions are encountered in Sphagnum bogs were previ-ously indicated (by sulfide and methane production, inaddition to methanogen detection) at depths greater than10 cm below the water surface [10, 16, 30, 37]. Sampleswere collected by inserting a closed-end, perforated pipe(1 in. diam.) into the mat. The sample was then collectedusing a hand pump, after flushing the length of plastictubing with ddH20. Samples ranged in texture andconsistency from moderately humic and watery, to thick

Table 1. Location and code of all bog sites in the study

Site Code Latitude Longitude Elevation (m asl)a Bog area (m2)

MassachusettsArcadia AR 42.31 72.43 95 1,190Bourne Hadley BH 42.56 72.09 274 105,369Black Pond BP 42.18 70.81 45 9,679Chokalog CHO 42.03 71.67 152 7,422Clayton CLA 42.05 73.30 210 73,120Hawley HAW 42.58 72.89 543 36,813Lily Pond Bog LIL 42.44 72.84 468 56,559Otis Bog OT 42.23 73.06 491 89,208Ponkapoag PON 42.19 71.10 47 491,189Quabbin QU 42.42 72.22 175 6,706Quag Pond QP 42.57 71.96 335 40,447Round Pond RP 42.17 72.70 78 10,511Snake Pond SNP 41.41 70.61 17Shankpainter SP 42.05 70.71 1 55,152Swift River SW 42.27 72.34 121 19,699Lake Jones WIN 42.69 72.00 323 84,235VermontCarmi CAR 44.95 72.90 133 38,023Chickering CHI 44.32 72.48 362 38,081Colchester COL 44.55 73.28 30 623,284Molly MOL 44.50 72.64 236 8,852Molly Soil SOIL 44.50 72.64 236Moose MOO 44.76 71.74 353 864,970Peacham PEA 44.29 72.24 468 576,732Snake Mt. SNM 44.06 73.27 313 248Springfield SP 43.33 72.51 158 435

am asl: meters above sea level.

S.E. MORALES ET AL.: BACTERIAL COMMUNITIES IN BOGS 35

sludge. Surface samples tended to be less degraded andlighter in color with intact Sphagnum particulate, where-as 1 m samples were generally moderately to completelydegraded, with a dark color, and lower water content.Subsamples (20 mL) were decanted into smaller tubesand fixed in the field with glutaraldehyde (final concen-tration 2.5%) for microscopic counts. A soil sample fromoutside the wetland, classified as a hemist histosol, wasalso taken at one site. All samples were kept on ice in thefield, and stored at 4-C in the laboratory until DNAextraction, and all extractions were performed within48 h of collection. Archival material was stored at j20-C.

Bacterial Abundance. We analyzed 59 samples.Four samples were not countable and were thereforeomitted from the study, leaving a set of 28 surface and 27one-meter samples. Prior to staining, we diluted fixedsamples 5- to 40-fold for optimum counting The dilutionwas then vortexed for 1 min, sonicated on a Branson5200 water bath sonicator for 1.5 min, and vortexedagain for an additional minute. Samples were thenfiltered through a 100-mm (pore size) nytex membrane(Sefar America, Kansas City, MO, USA) and 1 mL wasstained with 0.05% (v/v) SYBR Green II (MolecularProbes) for 15 min at 4-C. Stained preparations werethen collected on 0.2-mm (pore size) black polycarbonatefilters and mounted following the protocol of Fisheret al. [14]. Cell counts were measured at 1000�magnification on a Whipple grid using a Nikon E400epifluorescence microscope. Ten randomly chosen fieldswere screened using two different filters (XF53 dualwavelength filter and XF100 from Omega Optical Inc.,Brattleboro, VT, USA) to distinguish bacteria frombackground fluorescence.

DNA Extractions. Biomass in samples was pelletedby centrifugation in a Sorvall RC-5B centrifuge at 10,000� gfor 10 min. Total community DNA was extracted from0.25 to 0.5 g of pellet material using the MoBio Ultracleansoil DNA extraction kit (MoBio, Solana Beach, CA, USA)using water in the final step. Nitrosomonas europaea (ATCC19718) genomic DNA was used to produce an internalstandard DNA (described below). N. europaea was grownon the ATCC-specified medium prior to genomic DNAextraction using a Wizard Genomic DNA purification kit(Promega, Madison, WI, USA).

PCR Amplification. Amplification of the 16SrRNA gene was performed with primers universal forthe domain Bacteria: 8-27F labeled with 60 carboxy-fluorescein (6-FAM) and 1392-1407R [2] for T-RFLP,and S-D-Bact-1327 [42] for the clone library. PCR wascarried out using 50 ng of DNA in a final volume of50 mL, for 30 cycles, as described in [2]. PCR products(5 mL) were run on 0.8% agarose gels to verify ampli-

fication. Triplicate reactions were pooled, and DNAconcentrated and cleaned using a QIAquick PCR purifi-cation kit (Qiagen, Valencia, CA, USA) prior to restrictiondigests.

An internal standard for T-RFLP analysis was madeby amplifying the amoA gene of N. europaea (AMOFGGGGHTTYTACTGGTGGT and AMOR CCCCTCKG-SAAAGCCTTCTTC) from genomic DNA using a tetra-chloro-6-carboxy-fluorescein (TET) labeled forwardprimer. PCR product was cleaned as described above.

Cloning of 16S rRNA Gene. A clone library wasconstructed using DNA extracted from Molly Bog sur-face and 16S rRNA gene amplified using unlabeled for-ward primer. The clone library was generated by ligatingPCR products into pGem-T-easy and transforming theminto E. coli JM109 competent cells (Promega) accordingto the manufacturer’s protocol. Transformants werepicked and inoculated into a 96-well microtiter platecontaining 100 mL LB with glycerol and ampicillin. Theplate was incubated overnight at 37-C, then frozen ondry ice and shipped to The Institute for GenomicResearch (TIGR; Rockville, MD, USA) for DNA extrac-tion and sequencing.

Restriction Digests and T-RFLPs. Restrictiondigests contained 5 U of enzyme, 5 mL of labeled, pu-rified PCR product, and 1 mL Nitrosomonas standard in a20-mL total volume. Three different restriction enzymeswere used separately (HhaI, HaeIII, and MspI). Reactionswere performed at 37-C for 2 h followed by an inactiva-tion step at 65-C for 10 min. To remove fluorescence-quenching salts, and to concentrate DNA, digests wereethanol precipitated along with Pellet Paint NF copreci-pitant (Novagen, Inc., Madison, WI, USA) according tothe manufacturer’s protocol. Digested DNA was resus-pended in 5 mL of water for fragment analysis. Onemicroliter of each T-RFLP digest was mixed with 0.8 mL500 bp N,N,N 0,N 0-tetramethyl-6-carboxyrhodamine(Tamra) standard (PE Biosystems, Foster City, CA,USA) and brought up to a 20-mL total volume indeionized formamide. Capillary electrophoresis was per-formed on an ABI Prism 310 Genetic Analyzer (PEBiosystems) and T-RFs were analyzed using Genescan(PE Biosystems).

TRFLP Analysis. Profiles were analyzed usingGenescan and Genotyper programs (PE Biosystems).Parameters were set to exclude peaks under 50fluorescent units and those smaller than 25 bp or largerthan 600 bp. Internal Nitrosomonas standards for allsamples were aligned using Genotyper and comparedmanually to assess loss during sample processing, diges-tion efficiency, and accuracy of size determinations. Loadswere considered equal if the standard peak was 3000

36 S.E. MORALES ET AL.: BACTERIAL COMMUNITIES IN BOGS

(T1000) fluorescent units. Size determinations wereconsidered accurate when within 0.5 bp of the predictedsize. Profiles whose total fluorescence was below 10,000units were rerun and aligned to verify peak reproduc-ibility. Peak heights for most peaks were kept in the 50–300 fluorescent unit range with an average total fluores-cence of 13,104 units. Profiles were exported and datawere normalized to total peak height. T-RF data wereexported and used to run the tRFLP Fragment SorterProgram (http://www.oardc.ohio–state.edu/trflpfragsort/default.htm) [26, 34]. The Fragment Sorter programtakes into account the potential variability of T-RFswithin individual taxa and matches sample data withthose in the RDPII database to make assignments. Thisstep reduced the apparent taxonomic complexity bycompiling lists of three-enzyme rRNA fingerprints (ribo-types) from the T-RF peak data. Data sets for surface,1 m, and total community were generated by combiningmultiple sample data and creating a three-enzymeribotype master list with all the unique ribotypes present.Once generated, this list was used to create a presence/absence matrix for all the samples in the list.

Statistical Analysis. Data were organized as abinary presence–absence matrix, with rows representingribotypes and columns representing sites. The entriesindicate the presence (1) or absence (0) of a particularribotype at a site. Canonical correspondence analysis(CCA) was used to analyze the presence/absence matricescreated using SAS Version 8.02 Statistical Software

(Cary, NC, USA). Correspondence scores were plottedby site, abbreviated as shown in Table 1, for the first twocanonical dimensions.

Phylogenetic Analysis. All 16S rRNA genesequences were imported into Sequencher Version 4.1.4(Gene Codes, Ann Arbor, MI, USA) and contigs werecreated using sequences from the forward and reverseprimers, eliminating primer and vector sequences. Forclones whose forward and reverse sequence reads did notoverlap, only the forward primer derived sequences wereused. All sequences were compared to each other usingthe dirty assembly algorithm and realigner feature with20% overlap and 90 minimum match percentage. Rare-faction curves were generated using Analytic Rarefactionversion 1.3 (http://www.uga.edu/

e

strata/software/).Fragsort, like other programs [23], uses the T-RF

data to search sequences in a database [generated fromthe ribosomal database project (RDP)]. Although theprogram created a list of possible matching organismsclassified to the phylum (and in many cases, genus) level,present in our samples, we compared the three-enzymeribotypes with our clone library (Table 5). In silico digestswere performed, using Sequencher, on all clones using allthree enzymes. Generated patterns were then comparedto observed TRFs found from the same site. Based onpublished data [22] and personal experience, we set adefault gate of T3 bp to take into consideration variationsin fragment length assignments for clone library sequen-ces. Searching was done manually, and in cases where

Figure 1. Bacterial abundance bysite as estimated from direct countsusing SYBR Green II.

S.E. MORALES ET AL.: BACTERIAL COMMUNITIES IN BOGS 37

two of the enzymes fell within the T3-bp gate and a thirdone fell outside by G2 bp, a wider gate was accepted.

Phylogenetic analyses were performed using a TIGRin-house pipeline (Dongying Wu, unpublished data) thatautomatically assigns taxonomic tags to environmentalsmall subunit rRNA (SSU rRNA) sequences based onneighbor-joining phylogenetic trees. The 16S rDNAsequences obtained in this study were searched against

the Ribosomal Database Project (RDP) rRNA databaseby BLASTN. Top matches, as well as selected matcheswithin different score ranges, were aligned to the querysequence using CLUSTALW, and the alignments auto-matically trimmed using a Perl script based on alignmentquality score for all the columns. A neighbor-joiningphylogenetic tree based on the automatically trimmedalignments was constructed, and the taxonomy of the

Table 2. Two-way ANOVA of bacterial counts from surface and 1-m samples taken at 22 New England bogs

Source df SS MS F ratio P value

Bog 21 5 � 1015 2 � 2014 1.172 0.360Depth 1 9 � 1014 9 � 1014 4.746 0.041Bog � Depth 21 4 � 1015 2 � 1014 29.926 G0.001Error 396 2.543 � 1015 6.42 � 1012

Total 439 1.222 � 1016

Bog and depth were random effects in the mode, and there was a significant Bog � Depth interaction.df : degrees of freedom; SS: sum of squares; MS: mean square.

Table 3. Taxonomic assignment for the most represented ribotypes at 1 m and surface

Ribotype fingerprint Percent a

HhaI b MspI b HaeIII b Taxonomic assignment 1 m Surface 1 m+ surface

60 148 203 Bacteria, Planctomycetes, Planctomycetacia,Planctomycetales, Planctomycetaceae, Pirellula

96 92 94

60 147 245 Bacteria, Firmicutes, Bacillales 85 7590 159 202 Bacteria, Proteobacteria, Deltaproteobacteria,

Myxococcales, Cystobacterineae, Myxococcaceae,Myxococcus

85 82

91 159 202 Bacteria, Proteobacteria, Alphaproteobacteria,environmental samples

88 85 84

60 449 201 Bacteria, Proteobacteria, environmental samples 8197 265 268 Bacteria, Proteobacteria, Epsilonproteobacteria,

Campylobacterales, Helicobacteraceae, Helicobacter81 75

97 266 269 Bacteria, Proteobacteria, Epsilonproteobacteria,Campylobacterales, Helicobacteraceae

81 75

60 159 202 Bacteria, Proteobacteria, Deltaproteobacteria,Myxococcales, Cystobacterineae

88 77 80

60 162 246 Bacteria, Proteobacteria, Deltaproteobacteria,environmental samples

80 77 78

60 438 192 Bacteria, Proteobacteria, Alphaproteobacteria 7792 161 404 Bacteria, Proteobacteria, Deltaproteobacteria,

environmental samples77

98 267 270 Bacteria, Proteobacteria, Epsilonproteobacteria,Campylobacterales, Helicobacteraceae, Helicobacter

77

98 268 271 Bacteria, Proteobacteria, Epsilonproteobacteria,Campylobacterales, Helicobacteraceae, Helicobacter

77

60 160 247 Bacteria, Firmicutes, Bacillales, Bacillaceae,Virgibacillus

92 78

60 160 246 Bacteria, Firmicutes, Bacillales, Bacillaceae,Virgibacillus

84 76

90 159 202 Bacteria, Proteobacteria, Deltaproteobacteria,Myxococcales, Cystobacterineae, Myxococcaceae,Myxococcus

84

94 163 202 Bacteria, Proteobacteria, Deltaproteobacteria,Desulfovibrionales, Desulfovibrionaceae,Desulfovibrio

80

All samples included in the analysis.a Percent was based on specific ribotypes found in individual sets.b TRF in bp.

38 S.E. MORALES ET AL.: BACTERIAL COMMUNITIES IN BOGS

query sequence mapped to the taxonomy of the closest-branching RDP sequence in the midpoint-rooted neigh-bor-joining tree. This first-pass assignment was used toconstruct a second, higher-resolution tree using sequen-ces related to the query sequence at the family level, andtwo nonfamily members were also included to serve asoutgroups for tree rooting. The final taxonomic mappingof the query sequence was provided by the closestneighbor in the second tree.

Chemical Analysis. Water samples for chemicalanalysis were filtered in the field through glass fiber filterpaper (Whatman GFF), preserved, and analyzed asfollows: a subsample for Al, Ca, Cu, Fe, K, Mg, Mn,Na, Si, and Zn was acidified with nitric acid (finalconcentration 0.1 mol Lj1) and analyzed by inductivelycoupled plasma-atomic emission spectrometry (ICP-AES) using standard techniques; a subsample forNH4-N and NO3-N was preserved with sulfuric acid (topH 2) and analyzed on a flow injection autoanalyzerusing the Cd reduction method for NO3-N and thesalicylate–nitroprusside method for NH4-N; a subsamplefor DOC and DON was frozen as soon as possible aftersampling, and subsequently thawed and analyzed usingthe persulfite oxidation method; a separate subsamplewas frozen for subsequent analysis for SO4 and Cl by ionchromatography, and soluble reactive P by the stannouschloride molybdate blue procedure. Plant samples weredried, ground and microwave-digested in nitric acidfollowed by ICP-AES determination of Ca, Mg, Fe, Cr,Cu, K, Al, Co, B, Cd, Mo, Mn, Na, Ni, P, Pb, S, and Zn.Plant C, H, and N were determined by elemental analy-zer. All analyses were performed using standard methodsand quality control, including blanks, duplicates, andNIST traceable standard materials.

Nucleotide Sequence Accession Numbers. Thepartial 16S rRNA sequences used in the analyses havebeen deposited in GenBank under accession numbersAY775458–AY775538.

Results

Bacterial Abundance in New England Bogs. Directcounts on fixed cells stained with SYBR green werebetween 106 and 107 with average counts of 4.6 � 106 forsurface samples and 7.0 � 106 for 1 m samples (Fig. 1).Two-way ANOVA revealed a significant Bog � Depthinteraction term (Fig. 1, Table 2), thus we analyzed eachfactor separately. Overall, there was significant variationin bacterial abundance across our 22 sites (F21,418 =12.572, P G 0.0001). Within a site, bacterial counts weresignificantly higher for 1 m samples than for surfacesamples (paired t test: mean difference = 2,798,847, t21 =2.053, P = 0.05). The surface samples also had a lower

overall variance than 1 m samples, which is noticeable forboth grouped and individual sites (Fig. 1).

T-RFLP Analysis of Bacterial Community Similarity

Among Bogs. Individual enzyme profiles for surfacesamples contained, on average, 55 different peaksthroughout all enzymes, whereas 1 m profiles averaged42. The three-enzyme presence/absence matrix generatedfrom site data using the tRFLP Fragment Sorter Programidentified 618 distinct three-enzyme ribotypes at 1 m,753 at the surface, and 817 total ribotypes. There were 64ribotypes unique to 1 m samples, 199 ribotypes uniqueto the surface samples, and 554 shared ribotypes. Therewere 148 ribotypes that were unique to only one bog

Figure 2. Canonical correspondence analysis of bog samples,based on ribotype data, showing clustering between samples of thesame depth. Filled symbols represent 1 m samples, open squaresrepresent surface samples. A. Full view. B. Detail.

S.E. MORALES ET AL.: BACTERIAL COMMUNITIES IN BOGS 39

location, averaging to approximately three unique ribo<types per site. In contrast, a single ribotype classified as aPlanctomycetes was present in 46 of 49 (94%) samples(Table 3). Ribotypes present in more than 75% of thesamples were classified based on their RDP library match(Table 3). In 1 m samples, four out of the eight mostcommon ribotypes were assigned as Deltaproteobacteria.In surface profiles, the two major groups wereProteobacteria, with half belonging to the Delta and theother to the Epsilon subdivision. When data for bothdepths were analyzed together, 10 ribotypes were presentin at least 75% of samples. Firmicutes and Deltapro-teobacteria assignments were most frequently representedwith three ribotypes each.

Statistical analysis of the ribotypes profiles showedthat within the surface zone, the first two canonicaldimensions (CC1 and CC2) described 19.72% of the dataset variance, with a large number of sites clusteringtogether. The close or tight clustering of all but two sites,Quag Pond (QP) and Round Pound (RP), indicates ahigh degree of similarity among bacterial communities inthe surface samples from different bogs. The first twocanonical dimensions for the 1 m samples accounted for19.27% of the data set variance, with one outlier,Chickering Bog (CHI). The remaining sites clusteredtogether, but were not as closely grouped as the surfacesamples. When all samples were analyzed together, thesurface and 1 m samples constituted distinct clusters

Table 4. Taxonomic assignment of ribotypes contributing to separation of outlier sites

Ribotype fingerprint

Taxonomic assignment SitesHhaIa MspIa HaeIII a

60 131 38 Bacteria, Proteobacteria, Alphaproteobacteria,environmental samples

QPs

80 437 38 Bacteria, Proteobacteria, Alphaproteobacteria,Rhodospirillales, Rhodospirillaceae, Rhodospirillum

QPs

81 162 38 Bacteria, Proteobacteria, Alphaproteobacteria,environmental samples

QPs

331 149 38 Bacteria, Proteobacteria, Alphaproteobacteria,Caulobacterales, Caulobacteraceae, Caulobacter

QPs

333 440 38 Bacteria, Proteobacteria, Alphaproteobacteria,Rhodospirillales, Rhodospirillaceae, Rhodospirillum

QPs

341 86 38 Bacteria, Proteobacteria, Alphaproteobacteria,Rhizobiales, Methylocystaceae, Terasakiella

QPs

341 448 38 Bacteria, Proteobacteria, Alphaproteobacteria,environmental samples

QPs

81 474 38 Bacteria, Bacteroidetes, Sphingobacteria,Sphingobacteriales, Flexibacteraceae, Flectobacillus

QPs

81 475 38 Bacteria, Bacteroidetes, Sphingobacteria,Sphingobacteriales, Flexibacteraceae, Flexibacter

QPs

66 487 216 Bacteria, Proteobacteria, Betaproteobacteria,Rhodocyclales, Rhodocyclaceae, Zoogloea,environmental samples

CHP1m

66 489 218 Bacteria, Proteobacteria, Betaproteobacteria,Rhodocyclales, Rhodocyclaceae, environmentalsamples

CHP1m

66 491 220 Bacteria, Proteobacteria, Betaproteobacteria,Burkholderiales, Oxalobacteraceae, Herbaspirillum

CHP1m

81 483 38 Bacteria, Proteobacteria, Deltaproteobacteria,Myxococcales, Nannocystinea, Nannocystaceae,Nannocystis

QPs

98 167 83 Bacteria, Proteobacteria, Deltaproteobacteria,Syntrophobacterales, Syntrophobacteraceae,Desulfacinum

CHP1m, CHI1m

149 289 199 Bacteria, Firmicutes, Clostridia, Clostridiales,Acidaminococcaceae, Acetonema

CHP1m

206 302 38 Bacteria, Proteobacteria, Gammaproteobacteria,Oceanospirillales, Oceanospirillum

QPs

210 460 38 Bacteria, Proteobacteria, Gammaproteobacteria,Xanthomonadales, Xanthomonadaceae,Stenotrophomonas

QPs

a TRF in bp.

Ribotype fingerprint

40 S.E. MORALES ET AL.: BACTERIAL COMMUNITIES IN BOGS

(Fig. 2B). Although the first two canonical dimensionsaccounted for only 13.7% of the data set variance,vertical zones could be almost completely separatedbased on their bacterial ribotype profiles. The majorityof this variance is accounted for by the separation of twosurface samples in CC1 (QP and RP); and two other 1 msites in CC2 [Chickering Pond (CHP) and Chickering(CHI)] (Fig. 2A). Even without these outliers, a detailedview of the plot (Fig. 2B) clearly shows the segregation ofthe microbial communities into two separate groups (1 mand surface).

Individual samples taken from a different area in thesame bog were typically very similar (PON and SPR) at1 m depth. Surface samples replicated in this mannerexhibited more unique communities from within thesame bog (CHO and PON), although they still clusteredwith their respective depth. The single soil sample,obtained near one of the bogs, clustered with bog surfacesamples.

Correlations between the bacterial communities andenvironmental data including water chemistry (pH,DOC, ammonium, DON, NO3-N, SO4, Cl, P-SR, Al,Ca, Cu, Fe, K, Mg, Mn, Na, Si, and Zn) and plantchemistry (C, H, N, Ca, Mg, Fe, Cr, Cu, K, Al, Co, B, Cd,Mo, Mn, Na, Ni, P, Pb, S, and Zn) for three differentspecies (pitcher plant, leather leaf, and Sphagnum) wereconducted. Vegetation data, from the bog and the sur-rounding forest, and geographical data including lati-tude, longitude, bog area, and elevation were also used incorrelation analyses [18]. None of the tested parametersshowed any significant correlation to specific ribotypes orgeneral community structure. The only link between

sample segregation and chemistry was Ca2+ levels. Twosites, Chickering (8.59 mg/L) and Colchester (8.24 mg/L),showed approximately 8-fold higher Ca2+ levels whencompared to the average of all other bogs [0.98 mg/L, SD =7 (excluding Chickering and Colchester)].

Based on the CCA plots generated (Fig. 2), ribotypescontributing to the majority of the variance among thedata were identified (Table 4). Canonical dimensionsshowing the greatest variance shift depending on the dataincluded. Limiting the set of ribotypes to those explain-ing the greatest variance among the complete data set,(i.e., those unique to Quag Pond and Chickering Bogsamples) still gave an extensive list of candidate taxa.Because of the questionable nature of inferring physiol-ogy based on such assignments, greater confidenceregarding habitat differences that explain the uniquenessof those bacterial communities may await direct charac-terization of metabolic potential. Nonetheless, this workprovides an initial characterization of bog bacterialecology that may be used in metering change andevaluating additional habitat characteristics for theirimpact on those systems.

Taxonomic Representation as Identified by 16S rDNA

Sequences. A total of 75 clones were obtained andused for sequence determination and phylogeneticanalysis. Ten sequences were identified as possiblechimeras and discarded. The remaining sequences weredominated by Proteobacteria (36 clones/55%): [Delta(18/27%), Alpha (9/14%), Beta (7/11%), Gamma (2/3%)], followed by Firmicutes and Acidobacteria withseven representative clones (11%) each (Fig. 3). Two

Figure 3. Phylogenetic representa-tion within the 16S clone library.Phylogenetic placement wasassigned based on neighbor-joiningphylogenetic trees and BLASTNagainst the Ribosomal DatabaseProject (RDP) rRNA database.

S.E. MORALES ET AL.: BACTERIAL COMMUNITIES IN BOGS 41

clones were identified as possible members of separatecandidate divisions (TM7 and WS3). Other sequenceswere identified as being part of deep-rooted lineages suchas Actinobacteria [12] and Planctomycetes [15, 36, 38].

A comparison of the clone library to the ribo-types taken from the same site yielded eight matchesdominated by Acidobacteria (Table 5). The phylum Acid-obacteria has commonly been identified from envi-ronmental DNA using rRNA targets [4, 25, 33]. Theecological roles of members of that phylum howeverremain mostly undescribed because of the lack of cul-tivated representatives.

Rarefaction curves were constructed based on oper-ational taxonomic unit (OTU) groups formed at 90%sequence similarity (Fig. 4). Another rarefaction curvewas generated by grouping clones based on their topBLAST match accession number. Clones sharing thehighest similarity to the same organism were groupedtogether under the same OTU. The curve generated fromthis data set correlated to the 90% similarity rarefactioncurve (Fig. 4). The rarefaction curve generated did notflatten toward an asymptote, suggesting an incompletecoverage of the richness within the community.

Discussion

We used rapid community fingerprinting to comparebacterial community structure in surface and subsurfacemicrohabitats of Sphagnum bogs throughout northernNew England. We first compared bacterial communitiesamong 24 Sphagnum bogs; we tested the hypothesis thatSphagnum bogs represented a sufficiently homogeneouscollection of environments that bacterial communitystructure would be relatively similar among severalindividual sites. We also compared communities betweenT

ab

le5.

T-R

FL

Pan

d16

SrR

NA

clo

ne

lib

rary

com

par

iso

n

Clo

ne

Tax

onom

icas

sign

men

t

Pre

dic

ted

TR

Fs

Obs

erve

dH

aeII

IT

RF

Obs

erve

dH

haI

TR

FO

bser

ved

Msp

IT

RF

Hae

III

Hh

aIM

spI

Size

Abu

nd

ance

aSi

zeA

bun

dan

cea

Size

Abu

nd

ance

a

63A

cid

ob

acte

ria,

Aci

do

bac

teri

a,A

cid

ob

acte

rial

es,

Aci

do

bac

teri

acea

e,A

cid

ob

acte

riu

m21

628

226

721

6.45

14.1

528

1.6

3.05

268.

4614

.56

213.

7515

.28

279.

553.

7626

6.33

18.7

421

Aci

do

bac

teri

a,A

cid

ob

acte

ria,

Aci

do

bac

teri

ales

,A

cid

ob

acte

riac

eae,

Aci

do

bac

teri

um

214

9326

521

6.45

14.1

593

.53

3.57

266.

3318

.74

213.

7515

.28

91.4

7.42

34D

elta

pro

teo

bac

teri

a,Sy

ntr

op

ho

bac

tera

les,

Syn

tro

ph

ob

acte

race

ae,

Syn

tro

ph

ob

acte

r17

597

166

178.

096.

497

.83

5.84

167.

94.

48

92V

erru

com

icro

bia

,V

erru

com

icro

bia

e,V

erru

com

icro

bia

les,

Ver

ruco

mic

rob

iace

ae,

Ver

ruco

mic

rob

ium

175

9716

617

8.09

6.4

97.8

35.

8416

7.9

4.48

31A

cid

ob

acte

ria,

Aci

do

bac

teri

a,A

cid

ob

acte

rial

es,

Aci

do

bac

teri

acea

e,H

olo

ph

aga

203

361

162

203.

12.

5836

4.23

2.18

163.

162.

3320

1.49

5.52

51C

and

idat

ed

ivis

ion

WS3

237

214

6623

7.45

3.2

213.

172.

0963

.93

2.48

23A

lph

apro

teo

bac

teri

a,R

hiz

ob

iale

s,M

eth

ylo

cyst

acea

e,M

eth

ylo

sin

us

203

6114

820

3.1

2.58

58.4

57.

7414

7.09

7.43

201.

495.

527

Del

tap

rote

ob

acte

ria,

Syn

tro

ph

ob

acte

rale

s,Sy

ntr

op

ho

bac

tera

ceae

,Sy

ntr

op

ho

bac

ter

263

100

169

263.

241.

5297

.83

5.84

167.

94.

48

Pre

dic

ted

TR

Fs

fro

ma

clo

ne

lib

rary

wer

eco

mp

ared

agai

nst

ob

serv

edT

RF

sfr

om

the

sam

esi

te.

aP

erce

nta

geo

fto

tal

flu

ore

scen

ce.

Figure 4. Rarefaction curve generated from the 16S rRNA clonelibrary. The number of different clones was determined byalignment based 90% sequence identity. The number of OTU’s isplotted versus the number of clones.

42 S.E. MORALES ET AL.: BACTERIAL COMMUNITIES IN BOGS

surface and subsurface samples; we expected to finddistinct communities in these two microhabitats becauseof anoxic conditions in the subsurface samples.

Although bogs are oligotrophic and acidic, overalldiversity was high, with 817 ribotypes detected. Therewere more unique ribotypes in surface vs subsurfacesamples (199 vs 64), but total bacterial counts werehigher for the subsurface samples (Fig. 1 and Table 2).Although no data are available to explain this increase inabundance with depth, an increase in bacterial numbersin anoxic zones could correlate with a decreased amountof competition or predation by organisms present in theaerated zones. Alternatively, as counts were normalizedbased on cells/mL of sample, higher counts might havebeen an effect of solids content.

Canonical correspondence analysis of the ribotypedata showed a clear separation of samples based onsampling depth but relatively poor separation of bogssampled at the same depth (Fig. 2). At this level ofdiscriminating power, all of the truly ombrotrophic bogssampled were fairly similar in their bacterial communitycomposition. Chickering bog, which has been classifiedelsewhere as a poor fen, was identified as an outlier basedon T-RFLP profiles. This was possibly correlated to increasedcalcium levels, a mineral nutrient distinguishing fens frombogs [30]. The identification of a second bog (Colchester)with increased Ca2+ levels, but which was not an outlier inthe CCA analysis, implies that Ca2+ levels are not solelyresponsible for the unique composition of Chickering Bog.However, analyses to identify other environmental param-eters driving the community structure did not reveal cor-relation with any of the other tested parameters.

Analysis of individual ribotypes recognized Deltap-roteobacteria as a key group in bog ecology, having thehighest number of representatives among the mostfrequently assigned or Bcore^ ribotypes (Table 3) andthe clone library. It was not possible to conclude relativedominance of individual taxa, however; diversity indicesrequire species abundance data not obtained by our useof T-RFLP [9] or limited clone library analyzed.

Because these Northeastern U.S. Sphagnum bogscontained similar bacterial communities (total varianceof 4%), the sequence library derived from a single surfacesample might yield information regarding Sphagnum bogbacterial community composition across the region, andperhaps over an even wider geographic range. Thirteenlibrary sequences gave BLAST hits with sequencesobtained from wetlands. Ten of the matching sequences,including a TM7 Candidate Division match, came from aforested wetland in Georgia with a pH range of 2–4 [3],whereas the remaining three were from a Sphagnum Bogin Siberia [35]. Clones were derived from Proteobacteria,Verrucomicrobia, Firmicutes, Acidobacteria, Chloroflex-ales, Spirochaetales, Actinobacteria, Planctomycetes aswell as two Candidate Divisions. The clones included a

high proportion of Proteobacteria, particularly those ofthe Delta subclass. Most inferences about physiologicalcapability based solely on rDNA sequence are tenuous,but the metabolic diversity of the Deltaproteobacterialeaves many potential ecological roles open; members ofthat subclass are capable of dissimilatory sulfate reduc-tion, iron reduction, fermentative metabolisms, evenpredatory behavior and developmental life cycles [6, 8,29, 32]. The identified Deltaproteobacteria sequences,however, were most similar to Syntrophobacterales orDesulfovibrionales sequences, two Orders with represen-tatives from the sulfate-reducing bacteria [5, 20, 28].Published data for Molly bog [31] indicates a directcorrelation between pitcher plant density and increasedsulfate concentration. Thus we may have biased thesampling to favor sulfate-dependent or -responsive taxa.

The taxonomic diversity suggested from this studyindicates that Sphagnum bog environments represent arich source of potentially novel organisms. We haveestablished the homogeneity between similar wetlandsand our capacity to separate distinct ones based on theirbacterial community. The communities described werevertically stratified, possibly indicating that uniqueaerobic and anaerobic taxa are present [24]. A core bogcommunity was supplemented in each site with uniqueribotypes observed in only that or a limited group ofsites. Further testing of community fingerprinting of bogecosystems may enable the use of microbial communitiesas indicators of environmental changes. Isolation of bac-teria from this environment by culturing would also allowfor correlation with nucleic acid techniques and moredetailed examination of potential ecological functions.

Acknowledgments

This work was supported by funding from NSF (EPS-0082977). The authors would like to thank Aaron M.Ellison for sharing data as well as his extensive knowledgeof bog ecology and geography. We are grateful to DonStratton for assistance with the statistical analysis andCarol Tudhope and Grant Henderson for samplepreparation and data organization. Kevin Penn isthanked for performing DNA sequencing and analysis.

References

1. Blackwood, CB, Marsh, TL, Kim, S-H, Paul, EA (2003) Terminalrestriction fragment length polymorphism data analysis forquantitative comparison of microbial communities. Appl EnvironMicrobiol 69: 926–932

2. Braker, G, Ayala del Rıo, HL, Devol, AH, Fesefeldt, A, Tiedje, JM(2001) Community structure of denitrifiers, bacteria and Archaeaalong redox gradients in Pacific Northwest marine sediments byterminal restriction fragment length polymorphism analysis ofamplified nitrite reductase (nirS) and 16S rRNA genes. ApplEnviron Microbiol 67: 1893–1901

S.E. MORALES ET AL.: BACTERIAL COMMUNITIES IN BOGS 43

3. Brofft, JE, McArthur, JV, Shimkets, LJ (2002) Recovery of novelbacterial diversity from a forested wetland impacted by reject coal.Environ Microbiol 4: 764–769

4. Buckley, DH, Schmidt, TM (2003) Diversity and dynamics ofmicrobial communities in soils from agro-ecosystems. EnvironMicrobiol 5: 441–452

5. Castro, HF, Williams, NH, Ogram, A (2000) Phylogeny of sulfate-reducing bacteria. FEMS Microbiol Ecol 31: 1–9

6. Cummings, DE, March, AW, Bostick, B, Spring, S, Jr, Caccavo, F,Fendorf, S, Rosenzweig, RF (2000) Evidence for microbial Fe(III)reduction in anoxic, mining-impacted lake sediments (Lake Coeurd’Alene, Idaho). Appl Environ Microbiol 66: 154–162

7. Dedysh, SN, Panikov, NS, Liesack, W, Großkopf, R, Zhou, J,Tiedje, JM (1998) Isolation of acidophilc methane-oxidizingbacteria from Northern peat wetlands. Science 282: 281–284

8. Dhillon, A, Teske, A, Dillon, J, Stahl, DA, Sogin, ML (2003)Molecular characterization of sulfate-reducing bacteria in theGuaymas basin. Appl Environ Microbiol 69: 2765–2772

9. Dunbar, JM, Ticknor, LO, Kuske, CR (2000) Assessment ofmicrobial diversity in four Southwestern United States soils by16S rRNA gene terminal restriction fragment analysis. ApplEnviron Microbiol 66: 2943–2950

10. Duval, B (2000) Methane production and release from two NewEngland peatlands. Int Microbiol 3: 89–95

11. Ellison, AM, Gotelli, NJ (2002) Nitrogen availability alters theexpression of carnivory in the northern pitcher plant, Sarraceniapurpuea. PNAS 99: 4409–4412

12. Embley, TM, Stackebrandt, E (1994) The molecular phylogeny andsystematics of the actinomycetes. Annu Rev Microbiol 48: 257–289

13. Farrish, KW, Grigal, DF (1988) Decomposition in an ombrotro-phic bog and a minerotrophic fen in Minnesota. Soil Sci 145:353–358

14. Fisher, MM, Graham, JM, Graham, LE (1998) Bacterial abundanceand activity across sites within two Northern Wisconsin Sphagnumbogs. Microb Ecol 36: 259–269

15. Fuerst, JA (1995) The Planctomycetes: emerging models formicrobial ecology, evolution and cell biology. Microbiology 141:1493–1506

16. Galand, PE, Fritze, H, Yrjala, K (2003) Microsite-dependantchanges in methanogenic populations in a boreal oligotrophicfen. Environ Microbiol 5: 1133–1143

17. Goodwin, S, Zeikus, G (1987) Ecophysiological adaptations ofanaerobic bacteria to low pH: analysis of anaerobic digestion inacidic bog sediments. Appl Environ Microbiol 53: 57–64

18. Gotelli, NJ, Ellison, AM (2002) Biogeography at a regional scale:determinants of ant species density in New England bogs andforests. Ecology 83: 1604–1609

19. Gsell, TC, Holben, WE, Ventullo, RM (1997) Characterization ofthe sediment bacterial community in groundwater discharge zonesof an Alkaline Fen: a seasonal study. Appl Environ Microbiol 63:3111–3118

20. Holt, JG, Krieg, NR, Sneath, PHA, Staley, JT, Williams, ST (eds)(2000) Bergey’s Manual of Determinative Bacteriology. Williamsand Wilkins, Baltimore, MD

21. Johnson, LC, Damman, WH, Malmer, N (1990) Sphagnummacrostructure as an indicator of decay and compaction in peatcores from an ombrotrophic South Swedish peat bog. J Ecol 78:633–647

22. Kaplan, CW, Kitts, CL (2003) Variation between observed andtrue terminal restriction fragment length is dependant on true TRFlength and purine content. J Microbiol Methods 54: 121–125

23. Kent, AD, Smith, DJ, Benson, BJ, Triplett, EW (2003) Web-basedphylogenetic assignment tool for analysis of terminal restrictionfragment length polymorphism profiles of microbial communities.Appl Environ Microbiol 69: 6768–6776

24. Koizumi, Y, Kojima, H, Oguri, K, Kitazato, H, Fukui, M (2004)Vertical and temporal shifts in microbial communities in the watercolumn and sediment of saline meromictic Lake Kaiike (Japan),asdetermined by a 16S rDNA-based analysis, and related tophysicochemical gradients. Environ Microbiol 6: 622–637

25. Kuske, CR, Ticknor, LO, Miller, ME, Dunbar, JM, Davis, JA,Barns, SM, Belnap, J (2002) Comparison of soil bacterialcommunities in rhizospheres of three plant species and theinterspaces in an arid grassland. Appl Environ Microbiol 68:1854–1863

26. LaMontagne, MG, Leifer, I, Bergmann, S, Van De Werfhorst, LC,Holden, PA (2004) Bacterial diversity in marine hydrocarbon seepsediments. Environ Microbiol 6: 799

27. Liu, W-T, Marsh, TL, Cheng, H, Forney, LJ (1997) Characteriza-tion of microbial diversity by determining terminal restrictionfragment length polymorphisms of genes encoding 16S rRNA.Appl Environ Microbiol 63: 4516–4522

28. Loy, A, Kusel, K, Lehner, A, Drake, HL, Wagner, M (2004)Microarray and functional gene analyses of sulfate-reducingprokaryotes in low-sulfate, acidic fens reveal coocurrence ofrecognized and novel lineages. Appl Environ Microbiol 70:6998–7009

29. Madigan, MT, Martinko, JM, Parker, J (2003) Brock Biology ofMicroorganisms. Prentice-Hall, Englewood, NJ

30. Mitsch, WJ, Gosselink, JG (2000) Wetlands. Wiley, New York31. Mouser, PJ, Hession, WC, Rizzo, DM, Gotelli, NJ (2005)

Hydrology and geostatistics of a Vermont, USA kettlehole peat-land. J Hydrol 301: 250–266

32. Rendulic, S, Jagtap, P, Rosinus, A, Eppinger, M, Baar, C, Lanz, C,Keller, H, Lambert, C, Evans, J, Goesmann, A, Meyer, F, Sockett,SC, Schuster, SC (2004) A predator unmasked: life cycle ofBdellovibrio bacteriovorus from a genomic perspective. Science303: 689–692

33. Sait, M, Hugenholtz, P, Janssen, PH (2002) Cultivation of globallydistributed soil bacteria from phylogenetic lineages previously onlydetected in cultivation-independent surveys. Environ Microbiol 4:654

34. Sciarini, SM, Michel, F (2002) tRFLP Fragment Sorter. [2.5b3].Computer Program

35. Sizova, MV, Panikov, NS, Tourova, TP, Flanagan, PW (2003)Isolation and characterization of oligotrophic acido-tolerantmethanogenic consortia from a Sphagnum peat bog. FEMSMicrobiol Ecol 45: 301–315

36. Strous, M, Kuenen, JG, Fuerst, JA, Wagner, M, Jetten, MS (2002)The anammox case—a new experimental manifesto for microbi-ological eco-physiology. Antonie Van Leeuwenhoek 81: 693–702

37. Upton, M, Hill, B, Edwards, C, Saunders, JR, Ritchie, DA, Lloyd, D(2000) Combined molecular ecological and confocal laser scanningmicroscopic analysis of peat bog methanogen populations. FEMSMicrobiol Lett 193: 275–281

38. van Niftrik, LA, Fuerst, JA, Damste, JSS, Kuenen, JG, Jetten, MSM,Strous, M (2004) The anammoxosome: an intracytoplasmiccompartment in anammox bacteria. FEMS Microbiol Lett 233:7–13

39. Verhoeven, JTA, Maltby, E, Schmitz, MB (1990) Nitrogen andphosphorous mineralization in fens and bogs. J Ecol 78: 713–726

40. Williams, RT, Crawford, RL (1983) Effects of various physiochem-ical factors on microbial activity in peatlands: aerobic biodegra-dative processes. Can J Microbiol 29: 1430–1437

41. Williams, RT, Crawford, RL (1983) Microbial diversity ofMinnesota peatlands. Microb Ecol 9: 201–214

42. Yamamoto, H, Hiraishi, A, Kato, K, Chiura, HX, Maki, Y,Shimizu, A (1998) Phylogenetic evidence for the existence ofnovel thermophilic bacteria in hot sulfur-turf microbial mats inJapan. Appl Environ Microbiol 64: 1680–1687

44 S.E. MORALES ET AL.: BACTERIAL COMMUNITIES IN BOGS